The paper discusses a suitable segmentation method for extraction of specific objects from Magnetic Resonance Imaging (MRI). A particular attention is paid to detection and extraction of articular tissues from knee images. This is a pressing issue for physicians because MRI reveals often damage to articular cartilage which is shown by a minor change in a brightness scale. The image segmentation can provide a detailed colour map which shows distribution of the tissue densities. This algorithm is based on detection of local extremes in histogram and uses a membership function in order to allocate each image density into an output set. Each such set is given a colour from a predefined colour spectrum. This procedure can easily differentiate between the tissue structures based on the tissue densities.